Quantifying the Makeup Effect in Female Faces and Its Applications for Age Estimation

Abstract

In this paper, a comprehensive statistical study of makeup effect on facial parts (skin, eyes, and lip) is conducted first. According to the statistical study, a method to detect whether makeup is applied or not based on input facial image is proposed, then the makeup effect is further quantified as Young Index (YI) for female age estimation. An age estimator with makeup effect considered is presented in this paper. Results from the experiments find that with the makeup effect considered, the method proposed in this paper can improve accuracy by 0.9-6.7% in CS (Cumulative Score) and 0.26-9.76 in MAE (Mean of Absolute Errors between the estimated age and the ground truth age labeled or acquired from the data) comparing with other age estimation methods.